Fifteen years ago, the idea of a mining company handing ownership of its critical water infrastructure to an outside investor was close to unthinkable. Water was too important to let anyone else control. Today it is becoming a standard consideration. Desalination plants, pipelines, and treatment facilities are increasingly built, owned, and operated by third parties, with the mine as a long-term customer under a take-or-pay contract rather than as the owner of the asset.
The model was adopted for good reasons, and the savings it delivers are real. But it is priced on a belief about how the risk has been divided, and the instrument that is supposed to divide it does less than the price assumes. This piece is about what that instrument covers, and what it quietly does not.
Why the model looks smart, and largely is
Move the water plant off the mine’s balance sheet and two things improve at once.
First, capex falls. The hundreds of millions required to build a desalination plant and its conveyance no longer sit in the mining project’s capital budget, which lifts the project’s net present value on that line alone.
Second, and more importantly, the water asset gets funded by a cheaper pool of capital. Infrastructure and water investors underwrite long-dated, contracted, utility-like cash flows, and they price that risk at a lower cost of capital than mining equity or mining project debt would demand. A take-or-pay contract is precisely the instrument that turns a water plant into the stable annuity those investors want to hold. The result is a lower all-in cost of water for the mine and an acceptable return for the investor.
Everyone is better off, on one assumption. The assumption is that the risk has been cleanly divided the way the structure implies.
The instrument the whole price rests on
That clean division is the work of one instrument: the ring-fence. The water asset is placed in a separate project company, and the ring-fence is what separates that company’s fate from the fate of its owner, its lenders, and its customer. The cheaper cost of capital is, in effect, the market pricing the belief that the ring-fence has sorted the risk into tidy boxes. The investor holds water-infrastructure risk. The mine holds its supply security through the contract. Each is insulated from the other.
Ring-fencing is a good instrument. But it is a private-law instrument, and it carries a precondition that rarely gets stated out loud. It divides risk cleanly between two private parties, in ordinary conditions, while the matter in dispute stays private and bilateral. The moment a third party, a regulator, or a change in the mine’s own status enters the picture, the ring-fence starts to do considerably less than the price assumes. The rest of this article tests it against exactly those moments.
Where the ring-fence holds
Start with the case the model is built around, because it is genuinely a good one.
Suppose the investor that owns the water asset suffers a financial failure somewhere else, unrelated to the plant, large enough to threaten the whole entity. If the project company is properly ring-fenced, the failure above it does not reach the asset. The plant keeps running. The owner’s stake becomes something its creditors pursue, and ownership eventually transfers to someone else, as ownership can and does in normal times. The water keeps flowing throughout.
This is a bilateral, private, financial matter, and it is precisely the situation the ring-fence was designed to handle. It handles it well. The trouble is that this is also the case everyone stress-tests and then generalises from, concluding that because the structure survives a clean financial failure it will survive anything. It will not, and the reason is that the next three situations are not bilateral, not purely private, or not purely financial.
Where a regulator enters
The ring-fence binds the parties who signed up to it. It does not bind a regulator. A regulator answers to statute and to the public interest, and it owes nothing to the contracts inside the project company.
Consider the case where the water asset itself causes serious harm. This is a hypothetical, and it is the one that most exposes the limit of the structure. The liability now sits inside the very entity that owns the plant and holds the permits. The ring-fence was built to stop the asset’s troubles from climbing up to the owner. It offers nothing against a problem lodged in the asset itself, because there is no separation to invoke. And the response comes from a regulator, who can intervene in or halt the plant’s operation. A step-in right hands you the keys to an asset you may be legally barred from running. A shutdown ordered in the public interest is a physical interruption of supply that no private contract can reverse. If the plant also serves a community, the public-interest duty points toward stopping it faster, and the mine’s industrial supply becomes collateral to that public decision.
There is a second, quieter way the regulator’s presence bites, and it applies to insolvency proceedings as much as to environmental ones. These processes run on their own clock. Between the event that triggers them and the resolution that settles who controls the asset, there is an interval, often measured in weeks or months. During that interval the contractual instruments are suspended. An insolvency stay can freeze the step-in clause at the exact moment you would want to use it. The statutory process overrides the private right for its full duration. And the mine’s need for water does not pause to wait for the process to conclude. It is immediate, every hour, throughout. The interval is not a separate risk so much as the shape that regulatory and insolvency involvement takes: a period in which the ring-fence and everything attached to it is legally on hold while the requirement it was meant to secure continues unabated.
Where the mine’s own status changes
The investor’s side of the structure rests on a different assumption, and an equally fragile one. The take-or-pay contract is treated as a durable annuity, but its durability depends entirely on the counterparty, and a mine is not a utility customer. Three ordinary events, none of them a catastrophe, undermine it.
- The first is a sale of the mine. When the investor underwrote the deal, the take-or-pay was backed by a tier-1 miner’s balance sheet, its diversification across many assets, its credit standing, and often an explicit parent guarantee. That covenant strength is much of why the cash flow could be priced as utility-like in the first place. A sale to a single-asset operator or a junior leaves the contract wording identical while hollowing out what stands behind it. The parent guarantee commonly falls away with the sale, and what remains is a promise from an entity whose only revenue is the one mine and whose ability to honour a fixed water cost through a downturn is a fraction of what was originally underwritten. The annuity looks the same on paper and is worth materially less. Through a transaction it does not control, the investor’s slice has silently absorbed the credit quality of whoever ends up owning the mine, which is neither the contract’s nor the ring-fence’s to govern. Majors sell non-core and ageing assets down the quality ladder as ordinary portfolio management, so this is a routine event, not an exceptional one.
- The second is loss of operating rights. If the mine loses a permit or a licence, through legal challenge, regulatory action, or community opposition, the take-or-pay counterparty still legally exists but has no operation and no revenue. The contract says it must pay whether or not it draws water. An entity with no producing mine and no cash flow is a promise on paper. The investor’s right to be paid meets a counterparty with nothing to pay from, and the water plant’s own lenders were counting on that payment stream.
- The third is a decision to stop operations. When metal prices fall below the cost of production, owners mothball mines and place them on care and maintenance. This is a rational, cyclical business decision, and the sector has watched it happen repeatedly. Take-or-pay is meant to protect the investor here, since the mine must pay even when it is not drawing water. But that protection is only as strong as the payer’s balance sheet in a downturn, which is when it is weakest. A mothballed mine has every incentive to invoke force majeure, to seek renegotiation, or to litigate an obligation to pay for water it is not using. The annuity is most likely to be challenged precisely when commodity markets are down and the investor’s other holdings may be stressed as well.
Across all three, the same point holds. The investor priced infrastructure risk and is in fact holding mining risk, because the offtake is only ever as good as the mine’s continuity, and mining continuity is exactly the thing that is not guaranteed.
Where the risk actually landed
Put the two sides together. The cheaper capital was priced on the belief that the ring-fence had cut the risk into a clean infrastructure slice for the investor and a clean supply-security slice for the mine. In every situation involving a third party, a regulator, or a change in the mine’s status, that cut fails to hold, and the risk turns out to be sitting with a party who priced it as covered.
The mine can face an immediate supply gap with no contractual cure, because the instruments meant to protect it are suspended for the duration of the process that matters. The investor can face a mining-risk annuity dressed as an infrastructure return, exposed to the sale, the permit, and the price cycle of a counterparty it does not control. Neither party is holding what it believed it was holding, and the difference did not disappear. It was priced out of the deal and left in the structure.
The question for the sector
Ring-fencing is a fine instrument for what it actually is, a bilateral private arrangement between two entities under ordinary conditions. The model has quietly asked it to do a larger job: to hold risk steady across regulators, acquirers, and commodity cycles, none of which it was built to bind. The saving is real, and so is the gap.
So the question is worth asking plainly, before the first real test rather than after:
When the cheaper capital was priced, whose risk was the ring-fence assumed to be carrying, and does that assumption survive the first time someone other than the two original parties walks into the room?